Building an AI-Driven 4-Layer Intelligence Architecture with Neo4j for OSINT Investigations

Session track: Modern Applications

Session time: 01:00 am - 01:30 am UTC November 13

Session description:

In this session, Lucas Matheus will present how he transformed fragmented open data and OSINT sources into actionable intelligence that contributed to a real federal investigation in Brazil. Attendees will discover a practical, production-grade approach to building AI-powered knowledge graphs for complex, high-stakes investigations. The background is clear: critical public information about criminal networks — particularly child exploitation and human trafficking — exists but remains scattered across unstructured documents, social media, news, and public records. Traditional methods fail to reveal hidden connections, slowing investigations and limiting transparency. In Brazil, regional inequalities and sophisticated trafficking routes make the challenge even greater. Lucas will walk you through the complete 4-layer intelligence architecture he designed and deployed: Layer 1: AI-driven ingestion and linguistic decoding of raw public data and coded language. Layer 2: Behavioral logic and Modus Operandi (MO) matching using pattern recognition. Layer 3: Knowledge Graph construction and analysis with Neo4j, where entities (persons, locations, organizations, events) and rich relationships are modeled to uncover previously invisible networks — including new links involving high-profile figures. Layer 4: Human-in-the-Loop (HITL) validation combining criminological expertise with AI outputs. Throughout the talk, you will see real examples of Neo4j schema design, key Cypher queries used for network traversal and pattern detection, entity resolution strategies, and Graph visualizations that turned raw data into clear intelligence. You will also learn how AI agents accelerated entity extraction, cross-referencing, and real-time lead validation. By the end of this session, you will leave with concrete, actionable insights on: Modeling real-world criminal and influence networks in Neo4j Integrating Large Language Models with Knowledge Graphs for OSINT Building scalable, auditable intelligence pipelines Applying these techniques responsibly in sensitive “AI for Good” scenarios Whether you are a developer, data engineer, or AI practitioner working with graphs, this session will show you how to move beyond prototypes and deliver tangible real-world impact using Neo4j and modern AI.